Flow measurement has been an essential requirement in many commercial and industrial applications; one of the urgent needs is the low-power flow-sensing node for the Internet of Things (IoT) in the smart Energy-efficient Building (EeB). With the mature CMOS fabrication techniques, the low-cost MEMS flow devices combined with the highly integrated circuits (ICs) and micro mechanical components are able to be carried out. In particular, the target of this thesis, thermal flow sensors do not require any moving parts that make them perhaps the easiest flow devices to be implemented in the CMOS process due to the structural and electronic simplicity.

This thesis mainly focuses on the systematic design and fabrication of the low-cost micromachined calorimetric flow sensors by using t...[ Read more ]

Flow measurement has been an essential requirement in many commercial and industrial applications; one of the urgent needs is the low-power flow-sensing node for the Internet of Things (IoT) in the smart Energy-efficient Building (EeB). With the mature CMOS fabrication techniques, the low-cost MEMS flow devices combined with the highly integrated circuits (ICs) and micro mechanical components are able to be carried out. In particular, the target of this thesis, thermal flow sensors do not require any moving parts that make them perhaps the easiest flow devices to be implemented in the CMOS process due to the structural and electronic simplicity.

This thesis mainly focuses on the systematic design and fabrication of the low-cost micromachined calorimetric flow sensors by using the CMOS-MEMS technology. Primarily, a unified design platform with the theoretical modeling and simulation was proposed to get a comprehensive deep study of the thermoresistive micro calorimetric flow (TMCF) sensor system. Wherein, a general 1D model of TMCF sensor for two types of packaging: the open-space type and the channel type was proposed. Besides, the equivalent circuit model (ECM) for the TMCF sensor system was built and validated. This proposed unified design platform could be effectively applied in the design and optimization of CMOS TMCF sensor system. Accordingly, two batches of TMCF sensors were successfully designed and fabricated with the commercial CMOS MEMS technology, respectively.

The batch of monolithically integrated TMCF sensors was fabricated by using a 0.35μm 2P4M CMOS MEMS process. The fabricated TMCF sensors achieved a highest normalized sensitivity of 230mV/(m/s)/mW with the bidirectional flow detection of nitrogen (-11 ~ 11 m/s), which was two orders of magnitude higher than the previous CMOS flow sensors. Besides, the responses of TMCF sensors predicted by the 1D model and CFD model were in good agreement with the experimental data. Remarkably, an advanced T²MCF sensor was achieved, where the common issue of inconsistent output in the thermal flow sensor due to the variation of ambient temperature T_a was compensated and minimized. The T²MCF sensor showed the excellent temperature insensitive output with the maximum normalized variation of 0.5% under the different T_a of 22°C ~ 48°C. Compared to the uncompensated counterpart (49%), the measuring accuracy and stability of T²MCF sensor were greatly improved. Furthermore, a flow regime map (protrusion d* vs. reduced chip Reynolds number Re*) under different channel aspect ratio Ar was constructed to serve as a useful guideline for designing a well-packaged, reliable and accurate TMCF sensor system.

Another wafer-level encapsulated TMCF sensors was designed and fabricated by using the proprietary InvenSense CMOS MEMS technology. Thereinto, the fabricated Mo TMCF sensor with the pulsed operated CTD mode achieved an excellent normalized sensitivity of 112.4μV/(m/s)/mW with the bidirectional detection of nitrogen flow (-26 ~ 26 m/s). Besides, the measured TMCF sensor response time (τ_max < 3.63ms) shows good agreement with the proposed theoretical thermal RC model. Remarkably, an integrated CMOS MEMS Mo flow sensor which demonstrates a very compact system on chip (SoC) that features a very low-power Current Feedback Instrumentation Amplifier (CFIA) as a front-end readout circuit was also implemented, where the measured sensitivity of this SoC flow sensor was 35mV/sccm. For the sake of higher-order sensor value, a compact and low-cost micro BTU (μBTU) sensor system prototype, including a Mo TMCF sensor and a Poly-Si RTD fused on the same chip, integrated with a digital signal processing (DSP) unit, was proposed. The μBTU system achieved a detection range of 1.5 ~ 15milli-BTU/h and a sensor system accuracy of less than 6%.

The performance achieved by these CMOS MEMS flow/energy sensors, including the high sensitivity, low power consumption, large flow range, and rapid response ability, combined with the high level of integration and low-cost makes them possible to deploy as the flow/thermal sensing nodes of Internet of Things (IoT) for the flow and energy monitoring in the smart EeB.